The Cellar

The Cellar (http://cellar.org/index.php)
-   Technology (http://cellar.org/forumdisplay.php?f=7)
-   -   Building a gaming PC (http://cellar.org/showthread.php?t=29600)

plthijinx 11-04-2013 12:25 PM

Building a gaming PC
 
So a guy I know on a forum in f/b wants to build a gaming computer. THIS is what he came up with. His budget is roughly $1000 bucks. I looked it over and thought that it would be a nice set up. I did recommend going with the best Video card he could buy as that's the big punch nowadays with gamers....

thoughts? comments? switch component "A" with component "B"?

Undertoad 11-04-2013 12:49 PM

That'll do just fine. I split the drive space between a smaller SSD (128GB) and a standard 1TB hard drive. That way I boot fast and have all the huge game content on the big drive.

glatt 11-04-2013 12:49 PM

You're an electrical engineer type guy, right? Do you need a power supply that massive? How do you calculate what sized PS you need? Seems like a $90, 650 watt PS might be overkill.


But to be honest, I have no information to back that up. It's just a gut feeling.

plthijinx 11-04-2013 01:00 PM

true, BUT you have to pay attention to the power requirements of the mother board too. you can over size it and at the same time under size it. best to go with what the manufacturer recommends. to be honest i didn't look that closely at. But also it's powering everything else too under the hood. a 650 should do fine.

Undertoad 11-04-2013 01:19 PM

It's the video card that consumes all the power. They are mighty beasts these days, requiring their own power connector. 650W is appropriate.

plthijinx 11-04-2013 01:23 PM

Cool, Thanks Toad!

glatt 11-04-2013 01:45 PM

Heh.

I went to the Asus site and looked up that video card. 225w

Somebody ought to invent some ductwork that goes from inside the case to your cold winter mouse hand to keep it warm. You could probably bake a cupcake inside that case. My sister's childhood Easy Bake oven used a 100 watt bulb for a heat source, I think.

plthijinx 11-04-2013 02:54 PM

lol, yeah Glatt, i hear ya!

225w is a nice chunk of change for that. hell i've put in industrial plant LED floodlighting that runs on less than half that each fixture :)

tw 11-05-2013 07:16 AM

Quote:

Originally Posted by glatt (Post 882465)
I went to the Asus site and looked up that video card. 225w

If his computer and motherboard are consuming 600 watts, then also install a slot to toast bread.

Computer assemblers typically have no hardware knowledge. So we simplify things to keep them off technical help lines. Most all computers consume at maximum 200 watts. And often average around 100. So we tell them it consumes 300 watts. Since a power supply must be selected by amps for each DC voltage, well, that is too complicated for computer assemblers. So we make sure each DC voltage can provide enough current by doubling what the computer really needs. Now we have told them they need a 600 watt supply.

Then penis size makes a more manly system when they get a 650 or 1000 watt supply. Many computer assemblers even believe the 1000 watt supply is more reliable - because its number is bigger. They believe because they feel.

All for a computer that mostly consumes 100 watts and rarely more than 200 watts.

To keep these computer assemblers off the help lines, we tell them they need 225 watts for a video card that really only consumes typically 80 watts with short bursts that might be as much as 120 watts.

How large are power supplies in responsibly designed system by engineers? View numbers for Dell and HP. Typically around 225 watts for a system that has more than enough power for everything you might add to it. But the important number was never a supply's watts. The important number was always amps for each DC voltage (ie 3.3, 5, 12V).

When did Lite-On start making DVD drives? This was an LED manufacturer.

glatt 11-05-2013 08:03 AM

Quote:

Originally Posted by tw (Post 882487)
To keep these computer assemblers off the help lines, we tell them ...

You make it sound like keeping customers from having to call a help line is a bad thing. Seems to me you should add up the peak values for all the components, and get something that is equal to that number. Since it's unlikely that you will find a power supply with exactly that number, you should buy the next size up. And when you are looking at the price tag of the next size up, maybe there is a sale or something on a PS a little bit bigger that makes it a little more attractive.

You don't design for the average, you design for the peak. Imagine if dams were built for average river flow instead of for hundred year floods.

lumberjim 11-05-2013 08:24 AM

PS could mean power supply OR penis size, you know. I used an 800. I must be compensating for my small Asian penis.

Undertoad 11-05-2013 08:52 AM

This thread is called Building a gaming PC. The HP/Dell engineered machines are absolutely meaningless on this thread.

The overall wattage number is just the added total of wattages on all rails. On gaming rigs the +12V rail is the most important, but power supply wattage doesn't break that down, so a higher wattage supply may still not provide enough power on that rail. Power supplies of 8 years ago aren't enough to run modern video cards. They draw huge amperage!

Power supplies are less efficient at 100%, do you care?

Now about upgrades. Going to overclock? Many gamers do! But never the HP/Dell engineered systems.

Going to RAID? Many gamers do! But never the HP/Dell engineered systems.

Going to have 8 sticks of memory? Many gamers do! But never the HP/Dell engineered systems.

Add memory, overclock, add a second high-end GPU and run in tandem with CLI, now you need twice the wattage on that +12V line and the 650W won't handle it.

Not only that but! As power supplies are used, they lose capacity due to electrolytic capacitor aging. What was an appropriate supply this year will likely have 20% less capacity next year.

Finally, the power supply companies lie about their numbers.

Hell, go with the 750W. You're not going to draw more wall power; actually probably less, as the unit will run more efficiently. And it may even be able to power next year's model GPUs.

lumberjim 11-05-2013 09:10 AM

And think of how big your penis will look!

glatt 11-05-2013 09:52 AM

And with all that extra heat pumping out, it won't be playing hiding turtle from the cold.

plthijinx 11-05-2013 06:00 PM

Quote:

Originally Posted by glatt (Post 882491)
*snip*......You don't design for the average, you design for the peak. Imagine if dams were built for average river flow instead of for hundred year floods.

in what i do, you size transformers and cable sizes for peak, if not more (aka. oversizing). say you have a 480v to 240/120v single phase 50kVA transformer. is that good enough? well you have to look at your loads to find out. yeah, i started out assbackwards on purpose because i've had clients say that's what they want without knowing their power consumption for that feed. same here. you have the power.....only how much power gets used is up to all the components and with all the components consumption added up wattwise is the ps then big enough?

unless of course you're LJ. then the component is big enough. the only person i know that can trump the dos XX most interesting man in the world is him.

lumberjim 11-05-2013 06:31 PM

http://cdn.meme.li/i/mjb3m.jpg

tw 11-06-2013 06:57 AM

Quote:

Originally Posted by glatt (Post 882491)
Seems to me you should add up the peak values for all the components, and get something that is equal to that number. Since it's unlikely that you will find a power supply with exactly that number, you should buy the next size up.

Which is what engineers at HP and Dell have done for decades. His 800 watt supply would never see a 400 watt load. Similar computers from those manufacturers only need 200+ watt supplies to power every future part. These numbers are easily calculated since every peripheral connector (ie PCI or video card slot) and other connectors are limited to so many amps. Add all up and lower wattage supplies selected by HP and Dell engineers are more that sufficient.

Getting a layman to add those numbers is virtually impossible for a majority. Simpler to tell him he needs a 600+ watt supply and sell him an 800 watt supply for a computer that never consumes even 300 watts.

Watts is a useless design number. Current (amps) for each DC voltage is relevant. But most computer assemblers (amazingly) cannot add those numbers. Easier is to sell supplies on watts.

Some power supplies are rated at their output power. Others rated for how much is consumed. So a 250 watt power supply from a major manufacturer might also be sold as a 350 watt supply to computer assemblers. Nobody lied. Just another characteristic played with numbers that explains why many are told to buy an 800 watt supply - when a 200+ watt supply is more than sufficient. If he really needed a supply anywhere that large, then his computer can also toast bread.

This game continues. For example, I recently purchased a used power supply to fix a computer. That supply marketed to computer assemblers as ATX was missing essential parts necessary to make it ATX compatible. No problem. Holes for missing parts were occupied by jumpers. Parts removed from a defective supply replaced those jumpers to upgrade the new supply with ATX required functions. Supplies missing essential functions are extremely profitable since so many assume all ATX functions exist only if a computer boots.

Another feature of the game. Legally, only a computer assembler must know and meet industry standards and Federal regulations. If a power supply is missing those functions, then the computer assembler (not power supply manufacturer) is legally responsible. Many ATX power supplies are missing essential features that, for example, mean a power supply failure cannot and will not harm other computer parts. Then naive consumers actually believe a motherboard or disk drive damaged by a failing supply is acceptable. ATX standards define functions that make damage impossible. But some supplies are marketed missing such functions. Especially easy when the supply does not come with that long list of specs.

Step one for selecting any supply is a long list of numeric specs. Irrelevant is even if a consumer does not know what those numbers mean. If that long list of specifications is not provided, then a supply can be and may be missing essential and required ATX functions. And the few who actually know what those numbers mean cannot 'blow the whistle'. Unfortunately, most computer assemblers see a computer boot. That proves all ATX functions exist? No. But most consumers assume otherwise only because it boots.

An industry phrase (forgot the symbol) is now marketed that says a supply meets ATX standards - that is not missing essential functions.

Undertoad 11-06-2013 07:46 AM

Quote:

Getting a layman to add those numbers is virtually impossible for a majority. Simpler to tell him he needs a 600+ watt supply and sell him an 800 watt supply for a computer that never consumes even 300 watts.
No adding required, just go here, or to any other the other power calculators on the net, googlable with the phrase "pc power supply calculator".

Putting in the chosen components, allowing 3 high-performance fans and 3 USB power drawn devices (conservative in this day of USB charging), and with a 20% figure for capacitor aging, and overclocking a bit, the Thermaltake calculator says 476W is advised. We are giving headroom for additional components/USB charging and for enough amperage on the +12V rails.

Y'ever look at a video card in the last five years tw? They are massive. Take up more space than any other component and have their own dedicated rail. They are often as powerful and have as much memory as the motherboard. That's why it's a gaming rig, the type that HP/Dell generally do not sell.

tw 11-06-2013 08:36 AM

Quote:

Originally Posted by Undertoad (Post 882560)
No adding required, just go here, or to any other the other power calculators on the net, googlable with the phrase "pc power supply calculator".

A perfect example of why computer assemblers need 600+ watt power supplies for a system that only consumes 200+ watts.

We know that IDE disk drives can be powered from a USB port. That means a drive cannot draw more than 2.5 watts maximum. I added one IDE disk drive to the calculator. Not even an IDE interface card. Wattage increased from 88 to 105 watts. Had I added the interface card, it might have added another 14 watts. It says my disk drive consumes 17 watts? A 5400 RPM disk must not even consume 2.5 watt? How can this be if the calculator is so responsible and useful?

A disk drive interface card does not consumer 14 watts. It does not even consume half that.

The calculator also says these are continuously available - not maximum - numbers. So a disk drive that at maximum can only consume 2.5 watts may (according to the calculator) consume more than 17 watts? And its interface also is a toasty 15 watts? Yes. That is what the calculator says. Overly excessive numbers so that the power supply will be grossly more than required.

Thermaltake is selling power supplies to computer assemblers whose eyes glaze over with numbers. Your calculator says a less than 2.5 watt drive consumes more than 17 watts. Another example of why computer assemblers recommend 800 watt supplies for systems that barely consume 200 watts.

But again, wattage is useless for defining power requirements. Current draw for each voltage is relevant. Since consumers and that calculator cannot do that, then all simply provides a wattage numbers that is twice (or six times) more than required.

How many watts does the gaming video card consume? Telling us that the number is big is classic junk science. What exactly are the numbers for video cards in the past five years? As I say often and bluntly, useless claims have no numbers. Informed claims also provide perspective. How much current for each voltage was and is now required for video cards?

Demonstrated by the power supply is why so many recommendations for computers are often uninformed. How to separate the useful recommendations from hearsay: subjective and bogus claims provide no numbers. Or hype numbers from hearsay without any reasons to believe those numbers. Due to so much knowledge from hearsay, the same computer using a 200+ supply provided by engineers is also found with an 800 watt supply recommended by computer assemblers informed by a Thermaltake calculator. A majority will recommend an 800 watt supply. And other unnecessary components.

glatt 11-06-2013 09:03 AM

Quote:

How many watts does the gaming video card consume? Telling us that the number is big is classic junk science. What exactly are the numbers for video cards in the past five years. Wild speculation and junk science says it is large - a useless claim because it is subjective. Honesty demands that claim include a number.
I thought we covered this.

http://www.asus.com/Graphics_Cards/G...specifications

"Power Consumption:
up to 225W2 additional 6 pin PCIe power required"

If you don't trust the manufacturer, then how are you going to get that number? Pay for the thing and then test it? Ain't nobody got time for that.

tw 11-06-2013 09:19 AM

Quote:

Originally Posted by glatt (Post 882572)
I thought we covered this.
"Power Consumption:
up to 225W2 additional 6 pin PCIe power required"

That is not the power it consumes. That is an inflated number that customers are told. If the spec was written for a designer, it lists amperes for each voltage. Why is nobody's computer toasting bread? That question alone demonstrates wide variation between what is actually needed verses what a majority recommend.

The card may consume just over 100 watts. So we tell the computer assembler to provide a supply with an additional 225 watts. An example of why reality and what most computer assemblers recommend often varies significantly for other computer parts. We can also discuss another popular myth associated with Arctic Silver.

lumberjim 11-06-2013 09:21 AM

Is there any reason NOT to go with a power supply that provides more than what's strictly needed? I chose the 800 on the recco of the guy at Tiger Direct. I had a power supply fail in a Dell computer once, and did not want a repeat of that experience. I may very well upgrade my rig in future. The price difference was not prohibitive as I recall. Why are we spending so much time on this? Just tw wanking about how stupid everyone else in the world is, as usual? Thought so.

glatt 11-06-2013 09:35 AM

Quote:

Originally Posted by tw (Post 882576)
That is not the power it consumes. That is an inflated number that customers are told.

So how, specifically, do you find the real number?

tw 11-06-2013 09:37 AM

Quote:

Originally Posted by glatt (Post 882572)
If you don't trust the manufacturer, then how are you going to get that number? Pay for the thing and then test it? Ain't nobody got time for that.

Nobody said the manufacturer is lying. He has simply added a safety margin due to who his market is.

One can easily learn what the numbers really are by measuring. So many did that to learn their computers mostly consume around 100 watts. Most computers rarely exceed 200 watts in full load operation. One actually got his gaming computer to consume 400 watts.
What is your desktop power usage while browsing these forums?
Quote:

41 watts idle - 109 watts running Prime95
Someone blamed the measuring device because numbers did not agree with popular hearsay:
I have proved Kill-A-Watt accuracy .. NOT GOOD.
Quote:

at idle I was sitting at 65W and it peaked at 102W booting up.
These people obtained real world numbers. And that is the point. They got facts. Whereas most who recommend computer parts only recite popular hearsay. Always take what the majority recommend with a grain of salt. Especially when claims are subjective - not tempered by numbers. Demonstrated by the PSU is why so many myths are so often believed - ie Airborne cures the common cold or Geritol for a longer life.

If that video controller spec is useful, then it cited current (amperes) for each volt. Computer assemblers are told to get a power supply at least twice as large as necessary. A 200 watt computer needs a 400 watt supply. To avoid clogging help lines. Thermaltake may then recommend a 600 watt supply. Then consumers know they need 800 watts.

tw 11-06-2013 09:42 AM

Quote:

Originally Posted by lumberjim (Post 882577)
I chose the 800 on the recco of the guy at Tiger Direct. I had a power supply fail in a Dell computer once, and did not want a repeat of that experience.

That is also another popular myth. More watts means a more reliable power supply. Not true. In fact many power supplies have less stress when operating at closer to 100% power. Some technology supplies are at greatest stress when at 50% rated power.

But hearsay (and no numbers) easily convinces many that more watts mean better reliability.

Undertoad 11-06-2013 10:12 AM

If you don't like Thermaltake, try any of the 30 other power calculators.

glatt 11-19-2013 10:26 AM

Quote:

Originally Posted by Undertoad (Post 882462)
It's the video card that consumes all the power. They are mighty beasts these days...

I just saw Philthy's picture of the beast on Facebook. You are absolutely right.

Undertoad 11-19-2013 11:16 AM

Of course, and did you ever see one of those in an HP/Dell configuration? You can't even get one of those into the flipping case!

tw 11-19-2013 09:11 PM

Quote:

Originally Posted by glatt (Post 883829)
I just saw Philthy's picture of the beast on Facebook.

Good. You have numbers. What are they?

Undertoad 11-19-2013 11:28 PM

The EVGA GeForce GTX 660 has a minimum requirement of 24 amps on its own 12V+ rail.


watts = amps x volts OR P=IE as I learned it

tw 11-19-2013 11:47 PM

Quote:

Originally Posted by Undertoad (Post 883903)
The EVGA GeForce GTX 660 has a minimum requirement of 24 amps on its own 12V+ rail.

No way the board is consuming almost 300 watts. You would burn skin touching the heat sink. When electronics was consuming that much power, well, that is when Dec VAX would not operate unless the motherboard was fully covered with water.

That's the spec number we give to naive consumers. An actual and useful number is measured. And would be significantly less.

Undertoad 11-20-2013 12:04 AM

I guess you could try touching the heat sink, except that it's protected by a housing cooled by dual 80mm fans. Meanwhile those are manufacturer requirements, it's no longer 1995, and the GPU is much more powerful than the CPU in gaming rigs.

lumberjim 11-20-2013 12:31 PM

I put liquid cooling in mine. Just because I think that's cool. My computer has a radiator.

tw 11-20-2013 09:48 PM

Quote:

Originally Posted by Undertoad (Post 883906)
Meanwhile those are manufacturer requirements, it's no longer 1995, and the GPU is much more powerful than the CPU in gaming rigs.

Which means transistors need many times less power. Any such unit radiating 300 watts typically required a bathtub of water. Why are computer cases cooled in a 100 degree F room by only one 80 mm fan? Because this is no longer 1995. Transistors consume much less energy; radiate less heat.

Your reasoning is subjective. Today, computers generate many times less heat with each transistor. An IC radiating 300 watts would be immersed in a bathtub of water (ie Dec VAX). Today, a GPU only needs a fan. Which means it is not radiating almost 300 watts. Note a difference. Also provided with a request for numbers are hard numbers. Because subjective reasoning is only speculation.

One can easily learn what the numbers are by measuring. So many did that to learn their computers mostly consume around 100 watts. Most computers rarely exceed 200 watts in full load operation. One actually got his gaming computer to consume 400 watts.
What is your desktop power usage while browsing these forums?

Again the question. If Philthy's picture of the beast on Facebook provided useful facts (not speculation), then it provided numbers. What are those numbers?

lumberjim 11-20-2013 10:50 PM

Don't you get bored of yourself? Ever?

tw 11-20-2013 10:58 PM

Quote:

Originally Posted by lumberjim (Post 884013)
Don't you get bored of yourself? Ever?

Why are you wasting bandwidth?

lumberjim 11-20-2013 11:32 PM

Right. Silly question. A=A

glatt 11-21-2013 08:08 AM

1 Attachment(s)
Quote:

Originally Posted by tw (Post 884004)
Again the question. If Philthy's picture of the beast on Facebook provided useful facts (not speculation), then it provided numbers. What are those numbers?

You have been given those manufacturer provided numbers three times by my count, and you rejected them each time because you think the manufacturer is lying. Apparently, you expect someone (Philthy?) to hook the thing up and measure it? Why would he do that just to satisfy your curiosity?

But do a Google search, and see what it turns up: GeForce GTX 660 Ti 3GB
Attachment 46067

From this chart and several other sites, it looks like it typically uses 150 watts, but needs to be able to use twice that when playing a graphics intensive video game.

lumberjim 11-21-2013 10:08 AM

But what are the numbers? You have numbers. Please post them.

tw 11-22-2013 12:09 AM

Quote:

Originally Posted by glatt (Post 884045)
You have been given those manufacturer provided numbers three times by my count, and you rejected them each time because you think the manufacturer is lying.

Where did I say the manufacturer was lying? When marketing to people who make recommendations without numbers, then we routinely double power requirements. Current on each voltage - not watts - are the relevant numbers. But computer assemblers typically cannot do such calculations.

Point was not specifically power requirements of a GPU. The point is that recommendations without numbers are bogus. Your chart with numbers is consistent with previous numbers provided by others. The entire system consumes somewhere between 140 and 370 watts. Meaning a typical system never needed the 600+ watt PSU that so many recommend only using hearsay.

The chart does not say how much power a GPU actually consumes. And does not say what current requirements are for each voltage. It does not provide numbers sufficient for design (ie current for each voltage). But it demonstrates that some latest GPUs can require significantly more power.

Your numbers do demonstrate something interesting and useful. I have never seen a computer consume almost 500 watts instantaneously. I would have to dig further, but connectors have current limits. Implying that a system with the last two GPUs might exceed ATX connector limits. Might.

The chart of numbers is more consistent with what others measured (in "What is your desktop power usage while browsing these forums?") A typical GPU does not consume almost 300 watts. A system does. Most every system does not need the 600 and 900 watt supplies that so many recommend.

We would tell computer assemblers using the Radon 6990 or GE Force 590 that they need 1000 watt supplies. Because they do not know how to select hardware based upon what is relevant - current for each voltage.

You could have posted something 100% accurate. But without the associated numbers, it would be a useless recommendation. We still do not have enough information for design purposes. But your numbers confirm computers are not consuming 600 and 900 watts. Are not so hot as to toast bread.

classicman 11-29-2013 11:12 PM

Quote:

Originally Posted by lumberjim (Post 884013)
Don't you get bored of yourself? Ever?

Nope, because he is , like Charlie Sheen , winning !


All times are GMT -5. The time now is 05:31 PM.

Powered by: vBulletin Version 3.8.1
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.